Generating, benchmarking and simulating production schedules: From formalisation to real problems

Production scheduling has attracted the interest of production economics communities for decades, but there is still a gap between academic research, real-world problems, operations research and simulation. Genetic Algorithms (GA) represent a technique that has already been applied to a variety of combinatorial problems. Simulation can be used to find a solution to problems through repetitive simulation runs or to prove a solution computed by an optimization algorithm. We will explain the application of two special GAs for job-shop and resource-constrained project scheduling problems trying to bridge the gap between problem solving by algorithm and by simulation. Possible goals for scheduling problems are to minimize the makespan of a production program or to increase the due-date reliability of jobs or possibly any goal which can be described in a mathematical expression. The approach focuses on integrating a GA into a commercial software product and verifying the results with simulation.